Run Terminal-Bench and other Harbor evals on Vercel Sandbox
First reported · 36d ago · 1 source
Crest summary — written by AI from the 1 source below; every figure is checked against them · model deepseek-chat
Summary
Vercel now supports running Harbor evaluations, including Terminal-Bench, on Vercel Sandbox. Harbor is the open-source harness behind Terminal-Bench and other benchmarks.
Key facts
- Vercel Sandbox can run Harbor evals as of 2026-09-17.
- Harbor is the open-source harness behind Terminal-Bench.
- Harbor's registry includes SWE-bench, tau3-bench and OSWorld.
Why it matters
Developers can run Terminal-Bench and other Harbor benchmarks directly on Vercel Sandbox, simplifying evaluation of coding agents and models.
Companies
Technologies
- Vercelservice
- Harborproduct
- OSWorldproduct
- SWE-benchstandard
- Terminal-Benchproduct
- tau3-benchproduct
Related on Crest
Matched by rule, not by AI: a repository linked in the coverage, or a technology or company in this event that Crest tracks.
Repositories
Original sources (1)
The report this event was built from. Crest of Tech keeps no article bodies: the full text is at each source.
- VercelOfficial · PrimaryRun Terminal-Bench and other Harbor evals on Vercel Sandbox · By Elisabeth RülkeRead at Vercel